7. Concluding remarks
We introduced in this paper several important contributions to the tourism literature. First, we estimated a more robust productivity index that accounts for unobserved heterogeneity as well as the classical endogeneity problem in the estimation of input distance functions. Second, we provided a richer decomposition of productivity growth into three important components (input change, output change and frontier change). Third, we derived both short term and long-term productivity measures, providing hence some richer information for policy formulation in the tourism industry. Fourth, we provided measures of efficiency for each tourism destination, and applied the new methods to a rich of sample of leading tourism destinations and provided aggregate and individual country results. As mentioned, most existing studies in the area have focused only on one destination, or specific regions within one specific destination. Fifth, and finally, we measured productivity for the first time in this area using the Bayesian approach. The advanced assumption we impose on our model gives rise to a complicated statistical estimation problem which can be addressed successfully via Bayesian methods based on Sequential Monte Carlo/Particle Filtering (SMC/PF). We tested the performance of the model across various priors and also tested whether the instruments we selected for the reduced form are strong enough and proper.
